The improved fuzzy C-means clustering based on simulated annealing algorithm was applied to the stored-grain image segmentation. The computation load which exists in view of the fuzzy C-means value cluster algorithm in big and easy to fall into was partially most superior, the introduction simulation annealing algorithm, this algorithm starts when found the most superior region by the quick speed, found the globally optimal solution finally, might raise the convergence rate and the computation precision. The Matlab simulation experiments demonstr...